cost efﬁciency and ﬁnancial situation of local governments

Revista de Contabilidad Spanish Accounting Review 22 (2) (2019) 129-144

REVISTA DE CONTABILIDAD

SPANISH ACCOUNTING REVIEW

revistas.um.es/rcsar

Cost efficiency and financial situation of local governments in the CanaryIsles during the recession

Diego Priora, Ignacio Martín-Pinillos-Castellanosb, Gemma Pérez-Lópezc, José L. Zafra-Gómezd

a) Universitat Autònoma de Barcelona, Spainb) Universidad de la Laguna, Spainc) Facultad de Ciencias Económicas y Empresariales, University of Granada, Spaind) Universidad de Granada, Spain

dCorresponding author.E-mail address: [email protected]

A R T I C L E I N F O

Article history:Received 21 March 2018Accepted 4 December 2018Available online 1 July 2019

JEL classification:

Keywords:EfficiencyFinancial conditionBorrowingLocal public sectorOrder-m

A B S T R A C T

This paper examines cost efficiency in the Spanish municipal public sector, in a specific geographic area (theCanary Isles, Spain), with respect to financial condition and different types of municipal debt. The studyfocuses on municipalities dependent on tourism and on the consequences to them of the Great Recession,doing so via a two-stage analysis. In the first, the order-m method is used to evaluate the cost efficiency of 77Canary Isles municipalities, for the period 2008-12. In the second stage, we examine the effect produced oncost efficiency by different types of borrowing (long term, short term, financial and commercial) togetherwith other financial, economic, political and quality variables, using the model developed by Simar andWilson (2007), based on a truncated bootstrap regression with panel data. Empirical analysis shows thatin times of crisis there is a significant relationship between the components of financial condition andcost efficiency. Inconclusion, municipal cost efficiency increases with commercial debt, but decreases withfinancial debt. Furthermore, certain socioeconomic variables affect the levels of cost efficiency.

©2019 ASEPUC. Published by EDITUM - Universidad de Murcia. This is an open access article under theCC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Códigos JEL:

Palabras clave:EficienciaCondición financieraEndeudamientoSector público localM-Order

Condición financiera y eficiencia en los gobiernos locales de las Islas Canariasen tiempos de crisis financiera.

R E S U M E N

Este trabajo examina la eficiencia de costes para el caso particular del sector público municipal español,concretamente, para el caso de los municipios canarios, poniendo especial énfasis en áreas relacionadascon la condición financiera y diferentes tipologías de deuda, teniendo en cuenta las características propiasque estos municipios poseen, como un índice turístico elevado. En primer lugar, se utiliza el método deorden m para evaluar la eficiencia de costes de 77 municipios de las Islas Canarias, para el período 2008-12.En la segunda etapa, examinamos el efecto producido en la eficiencia de costes por diferentes tipologíasde endeudamiento (largo plazo, corto plazo, financiero y comercial) junto con otras variables financieras,económicas, políticas y de calidad, utilizando el modelo desarrollado por Simar y Wilson (2007), basadoen una regresión truncada con datos de panel. Nuestro análisis empírico muestra que en tiempos decrisis existe una relación significativa entre los componentes de la condición financiera y la eficiencia decostes. En particular, concluye que mientras la eficiencia de los costes municipales aumenta con la deudacomercial, disminuye con la deuda financiera. Además, mostramos que ciertas variables socioeconómicasafectan los niveles de eficiencia de costes.

©2019 ASEPUC. Publicado por EDITUM - Universidad de Murcia. Este es un artículo Open Access bajo lalicencia CC BY-NC-ND (http://creativecommons.org/licenses/by-nc-nd/4.0/).

https://doi.org/10.6018/rcsar.376091©2019 ASEPUC. Published by EDITUM - Universidad de Murcia. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

https://doi.org/10.6018/rcsar.376091

revistas.um.es/rcsar

[email protected]

http://creativecommons.org/licenses/by-nc-nd/4.0/


https://doi.org/10.6018/rcsar.376091



130 D. Prior, I. Martín-Pinillos, G. Pérez-López, J.L. Zafra-Gómez / Revista de Contabilidad Spanish Accounting Review 22 (2)(2019) 129-144

Introduction

The recent economic and financial crisis forced local gov-ernments to seek a more efficient allocation of their re-sources, due to the need to provide the same level of ser-vices with increasingly limited means (López-Hernández etal., 2012). Among other causes, this situation of scarceresources has arisen from falling public administration rev-enues, the increased competences acquired in recent years(Balaguer-Coll, 2004), greater demand for public servicesand the enactment of legislation1 strengthening economic-financial control over local entities (as indicated in the pre-amble to Act 27/2013, of 27 December, on the Rationalisa-tion and Sustainability of Local Administration). This back-ground, and the fact that municipal government is the level ofpublic administration that is closest to the citizen, underliesacademic interest in analysing municipal management per-formance (Zafra-Gómez & Muñiz, 2010) and thus in evaluat-ing the factors that determine efficiency (Bosch et al., 2012),especially those related to sources of funding.

In this context, it is especially important to analyse pub-lic sector borrowing, as local managers might raise levels ofpublic debt in order to compensate for lags and gaps betweenmunicipal income flows and costs. Studies of efficiency andborrowing have hypothesised that there may be an inverserelation between these two elements due to the substitutioneffect of borrowing for own resources, in what has beentermed a fiscal illusion (Cabasés et al., 2007; Pérez-López etal., 2014), and to the impossibility of reducing public spend-ing (Worthington, 2000; Geys, 2006; Da Cruz & Marquez,2014).

In view of these considerations, one of the main object-ives of our paper is to analyse the different characteristics ofborrowing and their relationship with cost efficiency, an areathat we believe has been insufficiently addressed in previousresearch, with special emphasis on the detailed analysis of dif-ferent types of borrowing, distinguishing between short andlong-term borrowing, and commercial and non-commercialborrowing, seeking to determine whether their particularcharacteristics influence municipal cost efficiency, especiallyin municipalities dependent on tourism and taking into ac-count the impact of the Great Recession (the most recenttransboundary crisis) on levels of municipal efficiency andfinancial condition (Ansell et al., 2010; López-Hernández etal., 2017; Plata-Diaz et al., 2017). In relation to this studygoal, we propose a series of hypotheses on which to base ouranalysis of the impact made by certain economic, financialand political factors on cost efficiency.

A twofold approach is taken to achieve these goals: first,order-m frontiers are used to calculate cost efficiency. Thisconcept was introduced by Cazals et al. (2002) and laterused by Simar (2003), who used a non-parametric modelto address the main problems encountered in data envelop-ment analysis (DEA), a method that has been widely usedin recent years, although rarely for the purpose of evaluat-ing municipal efficiency (Balaguer-Coll et al., 2007; Pérez-López et al., 2015; Narbón-Perpignan & De Witte, 2018a). Inthe second phase of this study, the determinants of cost effi-ciency are analysed in accordance with the method proposed

1The general legislation on Spanish local government, particularly LocalGovernment Act 7/1985 of 2 April (LRBRL); the local government financialregulations in Spain, in particular the Local Government Finance Act, ap-proved by Royal Legislative Decree 2/2004, of 5 March 5 (TRLRHL); andthe regulations on budgetary stability applicable to local government, whichincludes, in general, Organic Act 2/2012, of 27 April, on Budgetary Stabil-ity and Financial Sustainability, and its subsequent amendment Organic Act4/2012, of 28 September (LEPSF).

by Simar and Wilson (2007), which offers more robust res-ults than other models such as ordinary least squares (OLS)or Tobit (Da Cruz & Marques, 2014), applying a truncatedbootstrap regression with panel data and fixed effects. Manyprevious studies of the influence of certain factors on cost effi-ciency have used Tobit models or bootstrap truncated regres-sion with cross section data. In our opinion, the approachadopted in the present study provides greater robustness tothe analysis of the relationships considered. In this empiricalstudy, the sample population consists of 77 municipalities inthe Canary Isles, Spain (accounting for 83.5% of the total mu-nicipalities in the archipelago), all of which have fewer than50,000 inhabitants.

The results obtained show that short-term borrowing, com-pliance with credit limits, the index of budget sustainabilityand fiscal pressure per capita all present a significant inverserelationship with cost efficiency. On the other hand, com-mercial borrowing and the index of public service quality arepositively related to cost efficiency. We also obtained evid-ence regarding the cost efficiency of other factors such as theexercise of municipal government by a progressive politicalparty, of higher levels of population density and the signific-ant presence of a population aged over 65 years. All thesecircumstances were associated with higher levels of cost effi-ciency.

This paper is structured as follows. After this introduction,Section 2 describes the theoretical framework employed, re-garding cost efficiency, financial condition and municipal bor-rowing, and sets out the hypotheses to be tested. Section 3provides a detailed description of the research method, thestudy sample and the input and output variables used to cal-culate cost efficiency, together with the variables used to ana-lyse the financial-condition cost-efficiency relationship. Sec-tion 4 presents the results obtained, which are then discussedin Section 5, followed by a summary of the main conclusionsdrawn.

Theoretical framework for local government efficiency

Cost efficiency and financial condition in municipal govern-ment

Cost efficiency and the factors that affect it have beenwidely considered in previous research (De Borger & Ker-stens, 1996a, 1996b; Prieto & Zofio, 2001; Balaguer-Coll,2004; Balaguer-Coll et al., 2007; Balaguer-Coll & Prior, 2009;Benito et al., 2010; Bisogno & Cuadrado-Ballesteros, 2018;Cuadrado-Ballesteros & Bisogno, 2018; Narbón-Perpiñá & DeWitte, 2018a;2018b) ), seeking to identify inefficient activit-ies and to establish models based on local authorities foundto be relatively efficient. However, a more detailed under-standing is needed of the relationship between cost efficiencyand relevant financial factors, measured according to theconcept of financial condition (Zafra-Gómez & Muñiz, 2010).Accordingly, the first step is to consider which concepts of ef-ficiency and financial conditionshould be used.

Public administrators must decide which services shouldbe provided to citizens, and the quantity and quality of theseservices. To enable them do so, the preferences of societymust be determined, by reference to the elements charac-terising the demand for municipal services (Balaguer-Coll etal., 2007; Zafra-Gómez & Muñiz, 2010, Pérez-López et al.,2015).

Various theoretical approaches have been developed toconsider the most appropriate response to citizens’ needsregarding the quantity and quality of goods and services

D. Prior, I. Martín-Pinillos, G. Pérez-López, J.L. Zafra-Gómez / Revista de Contabilidad Spanish Accounting Review 22 (2)(2019) 129-144 131

provided. In particular, within the framework of the Theoryof Local Public Expenditures (Tiebout, 1956), it has been ar-gued that if there were more competition between local juris-dictions, municipal services would be more efficient, i.e. gov-ernments would use their resources in a way that minimisedthe cost of providing local public services. On the other hand,according to the Median Voter theory, citizens decide thelevel of expenditure they require by voting accordingly, opt-ing for the candidate offering the greatest service efficiency(i.e. services provided with respect to taxes paid) (Congleton,2002). As difficulties can arise in identifying citizens’ pref-erences, the theoretical approach of structuralism advocatesmeasuring them through proxy variables such as the demo-graphic, socioeconomic and fiscal factors within a communitythat affect the provision of public services (Hammer & Green,1996).

Of the different types of efficiency – technical or productiveefficiency, allocative or cost efficiency, distributive efficiencyand dynamic efficiency (Andrews & Entwistle, 2013) – instudies of local government, that of allocative or cost effi-ciency is commonly used (Vanden Eeckaut et al., 1993; DeBorger & Kerstens, 1996; Zafra-Gómez & Muñiz, 2010), dueto the complexity of measuring economic efficiency, amongother reasons because production prices are often not avail-able and because outputs are (totally or partially) determ-ined externally (Balaguer-Coll & Prior, 2009; Pérez-López etal., 2015).

There are various reasons for wishing to determine overallefficiency rather than that of specific public services (Borgeet al., 2008; Bosch, 2012; Pérez-López et al., 2015). On theone hand, as observed by Giménez and Prior (2007), mu-nicipalities provide multiple and varied public services, withthe common objective of satisfying the needs of local citizens,and therefore it is appropriate to consider the efficiency ofthese public services as a whole. Furthermore, major diffi-culties may arise in selecting and quantifying for analysis theproducts and services offered by each municipality and in ob-taining information on each of the services they provide in-dependently (De Borger & Kerstens, 1996). Finally, the avail-ability of other types of information may make it more ad-visable to analyse overall efficiency, since it makes no senseto assess the efficiency of services independently when thevariables that will be used to analyse the factors refer to themunicipality at the overall level (Borge et al., 2008).

In performing this analysis, it is necessary to be aware ofthe characteristics of local government financing, and its re-lation to efficiency, a question that has been addressed invarious studies. Thus, Dijkgraaf et al. (2003) studied therelationship between efficiency and financial situation, andconcluded that when the latter is difficult this forces muni-cipal managers to improve service efficiency as a means ofobtaining cost savings. Similarly, Zafra-Gómez and Muñiz(2010) analysed the relationship between efficiency and fin-ancial structure, observing that the most efficient municipal-ities are not those which have the best financial structure, butare characterised as presenting a degree of relaxation in thesearch for better financial margins, precisely because they aremore cost efficient. Benito et al. (2010) studied the relation-ship between efficiency and the economic level of local gov-ernment, and reached the conclusion that there is no signific-ant relationship between the two. However, most of the cor-relations found were positive, which suggests there is sometype of relationship between economic level and efficiency.

To properly address the relationship between efficiencyand financial situation, we must define what is meant by thelatter. In this respect, the literature refers to several concepts,

often interchangeable, such as financial position or conditionand fiscal stress, crisis or disaster. Among these terms, thebroadest and most appropriate concept for evaluating the fin-ancial health of local government is that of financial condi-tion (Zafra-Gómez & López-Hernández, 2006; Zafra-Gómezet al., 2009a; Hendrick, 2011). Diverse theoretical frame-works have been proposed for defining financial performance(Hendrick, 2011; Turley et al., 2015). These are character-ised by combinations of elements that may differ in their de-nominations, but, in general, use common indicators of li-quidity, debt and budget surplus/deficit, among other para-meters (Zafra-Gómez et al., 2009; Hendrick, 2011). Reviewsof the literature in this field show that the concept of fin-ancial condition includes a greater variety of such elementswith which to analyse the financial health of local administra-tions (Zafra-Gómez et al., 2009a; Zafra-Gómez et al., 2009b;Hendrick, 2011; Cabaleiro et al., 2013).

According to the Governmental Accounting StandardsBoard2 (1987), financial condition can be defined as the abil-ity to provide services whilst complying with present and fu-ture obligations. For Balaguer-Coll (2002, p.31), financialcondition is the entity’s capacity to meet its financial obliga-tions, including the payment of interest and the repaymentof debt, and its capacity to provide the level and quality ofthe services required for the welfare of its citizens, at an ac-ceptable level of taxes. Hendrick (2011) defined financialcondition as the state of equilibrium that exists between thedifferent dimensions or components of the government’s fin-ancial sphere in relation to its spending, obligations, fiscalresources, income and internal resources. Other authors,such as Greenberg and Hiller (1995), CICA (1997), Zafra-Gómez et al. (2006, 2009a), Groves et al. (2003) and Wanget al. (2007), have all developed particular views of thisconcept. In line with previous approaches in this field (Zafra-Gómez et al., 2006, 2009a; Cabaleiro et al., 2013) our studysample was composed following the approach recommendedby CICA (1997, 2009), according to which proper measure-ment of financial condition should include a consideration ofsustainability, flexibility and vulnerability.

For this purpose, sustainability is defined as the organisa-tion’s capacity to maintain, promote and preserve the socialwelfare of its citizens through the use of the resources at itsdisposition (Carmeli, 2002; PriceWaterhouseCoopers, 2006;Zafra-Gómez et al., 2009; Atan et al., 2010; Cabaleiro etal. 2013; Government of Western Australia: Department ofLocal Government, 2013). Flexibility is taken to be the or-ganisation’s ability to respond to changing economic and fin-ancial circumstances, within the limits of its fiscal capacity;this ability will be reflected in its reactions to such changesvia increases in taxation levels, in public debt or in transfersreceived (Zafra-Gómez et al., 2009a, 2009b; Hendrick, 2011;Cabaleiro et al., 2013). Vulnerability is understood as the or-ganisation’s level of dependence on external funding in orderto maintain public spending levels, via transfers, subsidiesand grants (Carmeli, 2002; Atan et al., 2010; Cabaleiro etal., 2013; Government of Western Australia: Department ofLocal Government, 2013); Zafra-Gómez et al., 2009a, 2009b;Plata-Díaz et al., 2017).

2Established in 1984, the Governmental Accounting Standards Board(GASB) is the independent, private-sector organization based in Norwalk,Connecticut, that establishes accounting and financial reporting standardsfor U.S. state and local governments that follow Generally Accepted Account-ing Principles (GAAP) (https://gasb.org. Accessed: 3 July 2018).

https://gasb.org


Previous empirical evidence and its relation to the theoreticalframework

Previous researchers in this field have placed special em-phasis on the analysis of efficiency in local administrations,mainly through the use of non-parametric techniques bywhich inputs and outputs are determined. A unit of effi-ciency measurement is then configured, either using directindicators or proxies (De Borger et al., 1994; De Sousa &Stosic, 2005; Balaguer-Coll, 2004; Balaguer-Coll & Prior.,2009; Borge et al., 2008; Balaguer-Coll et al., 2007); Zafra-Gómez & Muñiz, 2010).

In analysing financial and fiscal factors, some authors havestudied the relationships between cost efficiency and vari-ables that represent indicators of the three aspects that definefinancial condition. Thus, Balaguer-Coll et al. (2007) repor-ted an inverse relationship between cost efficiency and theindex of budget sustainability. Vulnerability has been ana-lysed by studying the level of transfers, and in this respectGeys (2006) reported that higher transfers were related tohigher rates of efficiency. However, other authors have con-cluded that an increase in resources that are not obtained dir-ectly from citizens can provoke a ‘fiscal illusion’ effect, mak-ing municipal managers feel less pressured to achieve effi-ciency in the delivery of public services (Giménez & Prior,2007; Bosch et al., 2012). Flexibility can be analysed in termsof many variables, including debt, fiscal pressure and fiscalsurplus (Zafra-Gómez et al., 2009a, 2009b; Hendrick, 2011;Cabaleiro et al., 2013).

Many studies have reported finding a negative relationshipbetween debt and cost efficiency, explained by the presenceof financial costs that increase public sector spending, thusreducing cost efficiency (Worthington, 2000; Geys, 2006;Benito et al., 2010; Da Cruz and Marques, 2014; Corderoet al., 2017). According to the literature on short-term bor-rowing (Worthington, 2000; Geys, 2006), this is an explan-atory variable of cost efficiency, with which it is inversely re-lated, due to the increased borrowing incurred when finan-cial costs rise. However, an increase in long-term debt is asso-ciated with greater cost efficiency (Benito et al., 2010). Othervariables, too, have been considered in this context. Thus,Geys (2006) and Pérez-López et al. (2014) reported findinga direct relationship between net savings and cost efficiency.Other studies have analysed the influence of fiscal pressure inthis respect, producing conflicting results. Some papers haveconcluded that greater fiscal pressure may reduce cost effi-ciency, arguing that when local governments lose the powerto generate current revenues, they become less motivated tomanage their resources efficiently (De Borger et al., 1994;Balaguer-Coll, 2002; Giménez & Prior, 2007; Balaguer-Collet al., 2007; Balaguer-Coll & Prior, 2009; Bosch et al., 2012;Cuadrado-Ballesteros et al., 2013). Others, however, affirmthat high taxes can lead voters to require more control of pub-lic spending, and to urge public managers to improve costefficiency (De Borger & Kerstens, 1996; Balaguer-Coll et al.,2007; Benito et al., 2010).

Research hypotheses

According to the above considerations, the research hypo-theses should contain the three elements proposed for meas-uring financial condition – sustainability, flexibility and vul-nerability. The first of these is considered via the budget sus-tainability index, which reflects the relationship between thevolume of local government spending and that of budget rev-enues. The higher the value of this indicator, the greater the

budget deficit. Following Balaguer-Coll et al. (2007; p. 437),sustainability is expected to be inversely associated with ef-ficiency, because a situation of deficit may be considered torepresent a mismanagement of public resources and, there-fore, reduced cost efficiency. Accordingly, the first researchhypothesis is posed as follows:

Hypothesis 1: An increase in the budget deficit (sustainability)will reduce cost efficiency.

The second element used in this study to measure financialcondition is that of vulnerability, i.e. the ratio of transfers re-ceived to total income. A high value for this indicator meansthat a large proportion of local government revenue is notobtained directly from the municipal population. This situ-ation can give rise to the fiscal illusion that any expense canbe met because there is sufficient income, and so municipalmanagers could perceive less pressure to operate more effi-ciently (Giménez & Prior, 2007; Bosch et al., 2012). Otherauthors have associated the effect of fiscal illusion with lowerlevels of public interest in supervising the policies adoptedby local government, an indifference that ultimately resultsin decreased cost efficiency (De Borger & Kerstens, 1996;Balaguer-Coll, 2002; Balaguer-Coll et al., 2007; Borge et al.,2008; Balaguer-Coll & Prior, 2009).

On the other hand, Geys (2006) reported that the largerthe volume of government transfers received, the higher theindex of cost efficiency, arguing that subsidies are subject togreater supervision and control, which limits the wasteful ap-plication of these resources. As there is no specific empir-ical evidence of the impact of this variable on cost efficiency,the second hypothesis merely suggests that an increase in thevalue of transfers will affect cost efficiency, without specify-ing its sign.

Hypothesis 2: An increase in the level of transfers (vulnerabil-ity) will affect cost efficiency.

The third element defining financial condition is flexibility,which is analysed by disaggregating the concept into threeaspects: borrowing, the index of compliance with credit lim-its and fiscal pressure. Moreover, it is necessary to distin-guish between short and long-term borrowing, commercialborrowing and financial borrowing. The variable ‘net savings’is defined as the resources generated with the current budgetthat are really available to finance capital expenses, once theannual amortisation of the debt has been financed. A greaternet saving must result either from increased income or fromdecreased spending (Pérez-López et al., 2014), and the lowerthe spending, the higher the cost efficiency. In other words,an increase in net savings will result in decreased spendingand, hence, greater cost efficiency. Note that in Spain a pre-requisite for requesting long-term funding, according to localgovernment regulations3, is that the net savings balance forthe previous year must have been positive.

Hypothesis 3a: An increase in positive net savings will increasecost efficiency.

An increase in the second of these variables, fiscal pressureper capita, defined as the ratio of municipal tax revenueto the number of inhabitants in the municipality, can harmcost efficiency because the local government could becomeless competent to generate current income, as a result of

3Article 53 of Legislative Decree 2/2004, of 5 March, approving the Con-solidated Text of the Local Government Finance Act.


being less motivated to manage its resources efficiently (DeBorger et al., 1994; Balaguer-Coll, 2002; Giménez & Prior,2007; Balaguer-Coll et al., 2007; Balaguer-Coll & Prior, 2009;Bosch et al., 2012; Cuadrado-Ballesteros et al., 2013). Onthe other hand, high local taxes might increase voters’ de-mands for stricter control of public spending and thus spurpublic managers to greater cost efficiency (De Borger & Ker-stens, 1996; Balaguer-Coll et al., 2007; Benito et al., 2010).

Although there is no unanimity on this question in the liter-ature, the following hypothesis proposes that an increase intaxes in relation to the number of local inhabitants will havea negative impact on cost efficiency. Thus:

Hypothesis 3b: An increase in taxes per capita will reduce costefficiency.

As previously explained, in this paper the question of flexib-ility receives special consideration, and is examined in termsof different indicators and types of borrowing. The literatureis not unanimous on the influence of borrowing on cost effi-ciency (Balaguer-Coll et al., 2007; Benito et al., 2010). Thetheoretical argument most frequently used is to assume aninverse relationship between borrowing and efficiency, dueto the presence of financial costs that raise the level of pub-lic spending and thus reduce cost efficiency (Worthington,2000; Geys, 2006; Benito et al., 2010; Da Cruz & Marques,2014; Cordero et al., 2017). However, this conclusion mightbe distorted by the different types of government borrowingavailable, which could plausibly affect the above relationship.Therefore, one of the aims of this study is to disaggregate theborrowing variable according to the term of the debt and thenature of the financing entity.

Many previous studies have examined the relationshipbetween cost efficiency and financial debt, but in this respectwe should distinguish between long and short-term finan-cial borrowing, as each type has its own characteristics thatcan influence levels of government cost efficiency in differ-ent ways. We adopt the initial premise that long-term finan-cing can only be used for long-term investments and, in turn,short-term financing can only be used to cover cash shortfallsrelated to short-term debts. Indeed, these characteristics aretypical of local government funding (Pérez-López et al., 2013,2014). The view that short-term borrowing is an explanat-ory variable of cost efficiency and has an inverse relation-ship with this parameter may be justified by observing thatincreased borrowing takes place in response to rising finan-cial expenses, as observed by Worthington (2000) and Geys(2006), and in consequence, cost efficiency deteriorates.

In consequence, the following hypothesis is divided intotwo elements: the first, on the relationship between cost ef-ficiency and short-term borrowing, and the second, on its re-lationship with long-term borrowing.

Hypothesis 4a: Increased short-term financial debt is associ-ated with reduced municipal cost efficiency.

An increase in borrowing that is used to finance investmentsin public facilities, in other words, long-term debt, can raisecost efficiency (Benito et al., 2010). In this case, the im-provement in cost efficiency would be directly related to thevolume of municipal investment, and so the following hypo-thesis is proposed.

Hypothesis 4b: There is a positive relationship between costefficiency and long-term financial debt.

However, we must differentiate between these forms of bor-rowing and commercial debt, which may be related with costefficiency in a different way. This contrast would arise fromthe fact that commercial borrowing does not usually incurfinancial costs, and so more income would be available forapplication to municipal services. As costs would not therebybe increased, these funds could be used to improve cost ef-ficiency. To our knowledge, no prior research evidence hasbeen reported in this respect. Accordingly, the following hy-pothesis is tested.

Hypothesis 4c: There is a positive relationship between costefficiency and commercial debt.

In addition to the above types of borrowing, we examine in-dicators related to compliance with borrowing limits (Pérez-López et al., 2014). The indicator of the latter factor is takenas the ratio between outstanding debt and current municipalincome; thus, the lower the ratio, the further the municipalfinances will be from compliance with the 110% index, thelegal limit for long-term investment4. In other words, thehigher the value of this indicator, the fewer resources will beavailable to the municipality to devote to public services and,consequently, the less efficient it will be.

Although there is no empirical evidence of the relationshipbetween this variable and cost efficiency, from its own defini-tion, an increase in the value of the indicator for compliancewith borrowing limits is expected to reduce cost efficiency,thus:

Hypothesis 5: There is an inverse relationship between compli-ance with borrowing limits and cost efficiency.

Quality, political and socioeconomic factors explainingcost efficiency

Other factors that may influence local government cost ef-ficiency are those related to service quality, political aspectssuch as political orientation and type of municipal govern-ment, and socioeconomic factors such as absolute population,population density, size of population aged under 15 yearsand/or over 65 years, unemployment rate, economic activityand the local importance of tourism.

There is no concrete evidence of the relationship betweenthe quality of municipal services and cost efficiency(Balaguer-Coll & Prior, 2009). In principal, an increase inquality would imply higher spending and, therefore, that themunicipality would be less efficient. On the other hand, ac-cording to the theory of Total Quality Management, investingin quality ultimately brings about increased cost efficiency(Prior, 2006).

The studies carried out to determine how the ideologyof the governing party (i.e., the political condition) affectsmunicipal cost efficiency have produced conflicting results(Vanden Eeckaut et al., 1993; Benito & Bastida, 2004; Bastidaet al., 2009; Kalb et al., 2012). In some cases, governmentswith a conservative majority have been found to achievegreater cost efficiency (Borge et al., 2008; Kalb et al., 2012;Da Cruz & Marques, 2014), but in others government by pro-gressive parties has been associated with higher levels of mu-nicipal cost efficiency (De Borger et al., 1994; Benito et al.,2010; Geys et al., 2010).

4Article 53 of Legislative Decree 2/2004, of 5 March, approving the Con-solidated Text of the Local Government Finance Act, and Final Provision No.31 of Act 17/2012, of 27 December, on the General Budget of the State forthe year 2013.


In this context, it is also important to consider whether theparty that governs the municipality does so with an absolutemajority or in coalition. In the former case, the governmentshould have less difficulty in achieving more efficient man-agement than if a coalition with other parties has to be nego-tiated and maintained (Benito et al., 2018). Such a coalitiongovernment might incorporate different styles of manage-ment and even experience policy conflicts among the partiesresponsible for municipal government. Empirical evidenceshows that strong political leadership (Borge et al., 2008)and a high number of votes for the strongest party (Bruns &Himmler, 2011) both have a positive influence on municipalcost efficiency. In contrast, other studies have reported find-ing a relationship between political fragmentation and costefficiency (Vanden Eeckaut et al., 1993; Balaguer-Coll et al.,2007; Borge et al., 2008). According to Kalb et al. (2012),this relationship would be explained by the political controlexerted over government partners giving rise to more effi-cient management.

In relation to the local population, larger municipalitiesmight achieve greater cost efficiency in services and facilit-ies, thanks to economies of scale (De Borger et al., 1994;Giménez & Prior, 2007; Benito et al., 2010; Bruns & Himmler,2011), but would also have to respond to greater numbersof users and provide more services. This obligation wouldbe accentuated if, in addition, there were a high density ofpopulation, since the higher the population density, the lar-ger the population served by the available infrastructure andservices, and this could produce negative effects from dis-economies of agglomeration (Borowiecki, 2013). In fact, ourreview of previous empirical evidence shows that there is nogenerally accepted view on the relationship between popu-lation size and cost efficiency; some studies have reported apositive relationship in this respect (De Borger et al., 1994;De Borger & Kerstens, 1996; Giménez & Prior, 2007; Benitoet al., 2010; Bruns & Himmler, 2011) while others have ob-served the opposite (Dubin & Navarro, 1988; Giménez &Prior, 2007).

What is well established, though, is that the distributionand characteristics of the local population can affect muni-cipal cost efficiency. When there is a large population ofinhabitants aged over 65 years, this can have a positive ef-fect on local government cost efficiency, as these residentsare able to exercise greater control over local policies, due totheir lower opportunity cost in obtaining information (Boschet al., 2012). On the other hand, Da Cruz and Marques(2014) created an aging index, incorporating both older andyounger population groups (those aged over 65 and under 15years), and reported that both groups require greater spend-ing on public services. They concluded that the higher thisindex, the lower the level of municipal cost efficiency.

With respect to local unemployment, contradictory evid-ence has been reported. According to Sung (2007) and Pérez-López et al. (2015), a higher rate of unemployment is associ-ated with lower cost efficiency, since as the number of joblesspersons increases, fewer resources will be available to main-tain public services, and if the same quantity and level of ser-vices are to be provided, this could lead to cost inefficiency.On the other hand, Kalb et al. (2012) affirm that a higherunemployment rate will improve efficiency by reducing pres-sure on the quality of public services, thus resulting in lowercosts for the municipality.

Regarding economic activity, Balaguer-Coll (2002) con-cluded that a higher level of economic activity improves costefficiency; however, other studies, such as Giménez and Prior(2007) and Balaguer-Coll and Prior (2009), have analysed

this question and concur that in municipalities with higherlevels of economic or commercial activity, policymakers willbe exposed to greater pressure to manage local services effi-ciently.

To conclude this review of socioeconomic factors, it is alsoimportant to consider the effect of tourism on the municipal-ity. When this activity is significant, it requires a certain levelof services, and increased demand in this respect may lead togreater cost efficiency in municipal government (Diez-Ticio& Mancebón, 2002). In contrast, Giménez and Prior (2007),Bosch et al. (2012) and Da Cruz and Marques (2014) all ar-gue that an increase in the local index of tourist activity isassociated with higher costs in terms of congestion and theseasonality of services, and hence cost inefficiency.

Methodology

The research aims of this study are addressed by meansof a two-stage analysis, following Simar and Wilson (2007),Balaguer-Coll et al. (2007, 2009), Balaguer-Coll and Prior(2009), Bosch et al. (2012), Cuadrado-Ballesteros et al.(2013) and Pérez-López et al. (2015), among others. In thefirst stage, the cost efficiency of various local public servicesis measured for each municipality in the sample, by the order-m frontier method. In the second, a truncated regression ofpanel data with fixed effects is applied, following Prior et al.(2008), Pascual Arzoz et al. (2008), Bastida et al. (2013)and Pérez-López et al. (2015). This process reveals whichvariables exert most influence on municipal cost efficiency.

3.1. Sample selection

The study hypotheses were tested using a sample of muni-cipalities in the Canary Isles (Spain). These islands are com-monly excluded from efficiency studies due to their special,insular characteristics (Balaguer-Coll & Prior, 2009; Zafra-Gómez et al., 2009a). Nevertheless, it is essential to analysesuch a region in order to determine how dependence on tour-ism may affect the provision of local public services, a ques-tion that has received little previous research attention. Themunicipalities analysed are located within the autonomouscommunity (region) of the Canary Isles, which is the secondmost important in Spain for domestic tourism (accountingfor 17% of the national total). In turn, Spain was fourth-ranked in the world by number of foreign tourists in 2010.The present research contributes new knowledge on the im-pact made by the 2008 Great Recession on cost efficiency, aconsequence referred to in a previous study of transboundarycrises (Ansell et al., 2010). The 77 municipalities in the studysample represent 87.5% of the total number of municipalities(88) in the islands. These 77 in particular were included be-cause a more complete data set was available for them thanfor the others, with respect to the study period, thus enablingus to address one of the main study goals, that of breakingdown municipal borrowing into commercial and financialdebt.Specifically, the study sample contains data from mu-nicipalities with fewer than 50,000 inhabitants5. This char-acteristic arises from the fact that the data for the outputsof local public services were obtained from the Spanish Sur-vey of Local Equipment and Infrastructure (EIEL), participa-tion in which is not mandatory for municipalities with over

5Towns and cities with over 50,000 inhabitants were excluded becauseEIEL does not collect this information for large municipalities. This limit-ation is also found in studies by Balaguer-Coll et al. (2007); Zafra-Gómezand Muñiz (2010), Pérez-López et al. (2015), Pérez-López et al. (2016) andPérez-López et al. (2018) , among others.


50,000 inhabitants (which in the Canary Isles is the case ofeight municipalities). In addition, three municipalities thatwere required to provide this information (all in the island ofEl Hierro) did not in fact do so until 2013 and so these, too,were excluded from the sample.

3.2. The order-m method of measuring cost efficiency (Stage1).

Various analyses based on frontier methods have been con-ducted to evaluate the efficiency of the local public sector.Most of these studies focus on inputs, in the assumption thatoutputs are determined by the institutional context that uni-fies the goods and public services to be provided, althoughmeasures of output quality are often included (Cabasés et al.,2007). Balaguer-Coll et al. (2007), among others, pointedout that since in most cases outputs are totally or partiallydetermined by forces beyond the control of the local corpor-ation, it seems more logical to evaluate efficiency in terms ofminimising inputs.

Among the various types of frontier analysis that may beconsidered, many recent studies of local government effi-ciency have applied non-parametric models. However, thesemodels can present estimation problems since the efficiencyestimates obtained by data envelopment analysis (DEA) andFree Disposal Hull (FDH) methods are very directly correl-ated (Simar & Wilson, 2007). Moreover, both methods arevery sensitive to outliers and extreme values, since they incor-porate all the data points (Daraio & Simar, 2007; Balaguer-Coll et al., 2013); in addition, they are susceptible to meas-urement errors, because they assume the absence of stat-istical noise (De Witte & Marques, 2010). Kneip et al.(1998) and Simar and Wilson (2007) were among the firststudies that attempted to overcome the non-robustness ofnon-parametric methods (Cordero et al., 2013). In this re-spect, Cazals et al. (2002) and Simar (2003) introducedthe concept of order-m frontiers, suggesting that efficiencyshould be evaluated with respect to a partial frontier ratherthan a complete one. By repeating the evaluation of the sub-sets of observations taken at random from the sample, a ro-bust approach can be configured. Moreover, this approachmakes it possible to perform the statistical inference of theefficiency indicators calculated, something which traditionalnon-parametric models do not allow. The order-m frontiermethod, accepting the non-convexity of the FDH, resolvesthe presence of extreme values, allowing the location of ob-servations beyond the estimated efficiency frontier, and there-fore the borders do not include all the data (Simar & Wilson,2007).

The algorithm for calculating order-m efficiency coeffi-cients considers a positive fixed integer “m”, such that froma certain level of input (xo) and output (yo), the estimatedefines the expected maximum value of m random variables(Y1 ,..,Ym), extracted from the distribution of the output mat-rix Y, which satisfies the condition Ym ≤ yo (Balaguer-Coll etal., 2013). Therefore, the following steps should be taken tocalculate the order-m estimator (Daraio & Simar, 2007):

1. For a given level of yo, extract a random sample of sizem6 as a substitute for ysm, such that ysm ≤ yo.

2. Using the random sub-sample, estimate the efficiencycoefficient as.

6In the present case, by employing the order-m frontier approach, wework with a cluster value of 50 observations, that is, m = 50, since thesensitivity analysis performed, with different values of m, revealed a degreeof convergence in the results for this value.

3. Repeat steps 1 and 2, B times7. In each round, a coef-ficient of efficiency is estimated, such that by the end of theprocess we have B coefficients of efficiency ab

S (b=1;2; ...; B).4. Finally, the arithmetic mean of the estimated efficiency

coefficients is calculated as:

αmS =

1B

B∑b=1

abS

In the present study, the selection of study variables is basedon those used in previous empirical research (De Borger &Kerstens 1996; Balaguer-Coll et al., 2007; Giménez & Prior,2007; Balaguer-Coll & Prior, 2009; Benito et al., 2010; Geyset al., 2010; Da Cruz & Marques, 2014; Ibarzola et al., 2017),taking into account the availability of information.

The inputs selected were the amounts of the respectivechapters of the expenditure budget, following previous em-pirical studies of municipal cost efficiency such as those byZafra-Gómez and Muñiz (2010), Cuadrado et al. (2013) andPérez-López et al. (2015), considering current and capitalexpenses and excluding financial expenses (Chapters 1, 2,4, 6 and 7). Budget settlement data were included in orderto avoid the inclusion of provisional data, and consolidateddata were used, in order to avoid the possible masking ofdata through the use of instrumental agencies (Fresneda &Hernández, 2018). The information on inputs was obtainedfrom the website of the Ministry of Finance and Public Admin-istration, via the General Secretariat for Regional and LocalCoordination8.

To define the outputs, we analysed the competences as-signed to these municipalities under applicable legislation9,using proxies in order to avoid difficulties in directly eval-uating the product or service provided. Most of these mu-nicipal services are described by the indicators included inthe EIEL, as presented on the website of the Ministry of Fin-ance and Public Administration (State Secretariat for PublicAdministration) and in the Unifica internet application madeavailable by the Government of the Canary Isles. Therefore,the variables selected for each entity were the total budgetin euros, as the input measure, and the number of inhabit-ants, the length of the road network (m), the volume of wasteproduction (tons), the area of public facilities (sq. m.), thenumber of street lamps, the length of the water distributionnetwork (m) and the surface area of cemeteries (sq. m.), asthe output measures.

3.3. Truncated bootstrap regression with panel data (Stage2).

The research hypotheses were tested by applying the pro-posal of Simar and Wilson (2007) to the results obtained inthe first stage, using a truncated bootstrap regression withpanel data, with fixed effects. This method was chosen be-cause alternative procedures, such as OLS and Tobit mod-els, have been criticised in the sense that correlation prob-lems could arise between the first and second-stage variables,possibly invalidating the results of the statistical inference(Fernández-Santos et al., 2015). Furthermore, as observed

7Although in most applications it is reasonable to use B= 200 (Balaguer-Coll et al., 2013), in the present study B = 2,000 is assumed, as suggestedby De Witte and Geys (2013).

8http://serviciosweb.meh.es/apps/EntidadesLocales/ consulted inJune 2015. Except data from the municipality of Betancuria (Fuerteven-tura) for the year 2010, since this information was not on the Ministry’swebsite. Instead, it was obtained directly from the municipality.

9Article 26 LRBRL specifies the powers of the municipalities.

http://serviciosweb.meh.es/apps/EntidadesLocales/


Table 1Description of the variables included in Stage 2 10

19

Table 1. Description of the variables included in Stage 210

VARIABLE DEFINITION

Municipal cost efficiency (dependent variable)

Ratio of municipal spending to municipal services. Normally takes values between 0 and 1, but as the order-m method allows superefficiency values, values greater than one may be obtained.

Index of budget sustainability Ratio of expenditure (in chapters 1-7) and income (chapters 1-7).

Level of government transfers received Current and capital transfers from other public entities as a percentage of total income.

Short-term borrowing Short-term consolidated municipal debt.

Long-term borrowing Long-term consolidated municipal debt.

Financial debt (short and long term) Consolidated municipal debt owed to financial entities.

Commercial debt Consolidated municipal debt owed to budget and non-budget creditors and to other public entities.

Compliance with borrowing limits Compliance with the borrowing limits imposed by national regulations, i.e. <110% of outstanding debt, in relation to current income.

Net saving Current income less current expenses, less debt amortisation.

Fiscal pressure per capita Tax revenue in relation to the number of inhabitants.

Population density Surface area (km2) of the municipality in relation to the number of inhabitants.

Population tranche Categorical variable that classifies municipalities by population size: 0-5000, ppn=0; 5001-20000, ppn=1; 20001-50000, ppn=2.

Population aged under 15 years Number of inhabitants aged under 15 years in relation to the total number of inhabitants.

Population aged over 65 years Number of inhabitants aged over 65 years in relation to the total number of inhabitants.

Index of unemployment Number of persons unemployed in relation to the total number of inhabitants.

Index of economic activity Tax income from total business activities (industrial, commercial and services) and professional activities in relation to the national total and in relation to the total number of inhabitants of the municipality.

Index of tourist activity Tax income from economic activities in the tourism sector in relation to the national total and in relation to the total number of inhabitants of the municipality.

Index of average quality11 Variable measuring the average quality of local public services, each of which is assigned the value 1, 2 or 3, corresponding to an evaluation of Poor, Fair or Good quality, respec-tively. For refuse collection, the quality values used are Sufficient= 2 or Insufficient= 1.

Political orientation of municipal government

Variable assigned the following values, according to the orientation of the Mayor’s political party: Progressive=0; Conservative=1

Type of municipal government Variable assigned the following values, according to the type of governing majority of the Mayor’s political party: Absolute=0; Coalition=1.

Source: The authors.

10 Our analysis of financial condition does not include liquidity because when the study was performed the data for this variable were not available in the Local Information System of the Canary Isles Autonomous Community (Unifica, the official website of the Canary Isles regional government) for the municipalities analysed in this period, and its inclusion would mean losing part of the sample. 11The Quality Indicator is composed of the following quality indices obtained from the EIEL: condition of highway infrastructure, condition of the refuse collection network, condition of public equipment, condition of street lamps, condition of the water distribution network. In addition, equipment with the status “in the process of implementation” is excluded from the results since it is not currently being used and therefore is not providing any service to the local inhabitants. In the case of road infrastructure, in some cases the level of quality is stated as “unpaved” which we categorise as poor quality.

Source: The authors.

10 Our analysis of financial condition does not include liquidity because when the study was performed the data for this variable were not available in the Local Information Systemof the Canary Isles Autonomous Community (Unifica, the official website of the Canary Isles regional government) for the municipalities analysed in this period, and its inclusionwould mean losing part of the sample.

11 The Quality Indicator is composed of the following quality indices obtained from the EIEL: condition of highway infrastructure, condition of the refuse collection network,condition of public equipment, condition of street lamps, condition of the water distribution network. In addition, equipment with the status "in the process of implementation" isexcluded from the results since it is not currently being used and therefore is not providing any service to the local inhabitants. In the case of road infrastructure, in some cases thelevel of quality is stated as "unpaved" which we categorise as poor quality.


by Balaguer-Coll et al. (2007), many two-stage studies areinconsistent because they assess efficiency (in the first stage)using non-parametric analysis, but consider the explanatoryfactors in the (second stage) by a parametric method.

The general specification of the truncated regressionmodel used in our approach is described by the followingequation:

αmSi = a + β1 Z1,i t + β2Z2,i t + β3Z3,i t + β4Z4,i t + ϵi

where:

i = 1, 2, . . . ,77 (municipalities)

t = 2008, . . . , 2012 (years)

αmSi is the dependent variable represented by the ef-

ficiency obtained for each municipality (obtainedin the first stage).

a is a constant term.

Z1,it is the vector of the financial explanatory vari-ables.

Z2,it is the vector of the political explanatory vari-ables.

Z3,it is the explanatory variable of service quality.

Z4,it is the vector of the socioeconomic explanatoryvariables.

β1,...,4 are the vectors of the parameters to be estim-ated in the second stage, which determine the re-lationship between the independent variables andthe dependent variable.

ϵi is the independent error term (or statisticalnoise), normally-distributed N (0, δ2

ϵ , and left trun-cated at (1- bβ Zi).

Table 1 details the second stage variables used in the study.These were chosen taking into account the methods adoptedin previous empirical studies in this field, such as De Bor-ger et al. (1994), Balaguer-Coll et al. (2007), Balaguer-Colland Prior (2009), Benito et al. (2010), Bosch et al. (2012)and Pérez-López et al. (2015), and according to the avail-ability of information and the legal framework applicable tothese local entities. The information related to these vari-ables was obtained from the following sources: the Ministryof Finance and Public Administrations (Secretariat for Re-gional and Local Administration and the Virtual Office for theFinancial Coordination of Local Entities); the National Insti-tute of Statistics (INE); the Survey of Local Equipment andInfrastructure (EIEL); the Unifica website of the AutonomousGovernment of the Canary Isles; the Economic Yearbook pub-lished by La Caixa; the website of the Ministry of the Interior;and the websites of relevant political parties.

Analysis of results: relationship between cost efficiency,financial condition and debt

This section presents the results obtained in the first andsecond stages, which are used to test the research hypo-theses.

Table 2 shows the descriptive statistics of the variables in-cluded in the first stage of the analysis.

The descriptive statistics for each year of the first-stagevariables are shown in Annex 1. Notably, the average total

Table 2Descriptive statistics: input and output variables (2008-2012)

20

4. Analysis of results: relationship between cost efficiency, financial condition and debt

This section presents the results obtained in the first and second stages, which are used to test

the research hypotheses.

Table 2 shows the descriptive statistics of the variables included in the first stage of the

analysis.

Table 2. Descriptive statistics: input and output variables (2008-2012).

Variable Mean Median Standard deviation Maximum Minimum

Total cost (euros) 13,388,847 8,643,118 12,451,851 77,448,391 1,279,124 No. of inhabitants (n) 12,865 8,605 11,085 44,007 772 Length of street network (m) 113,097 92,895 73,042 309,058 11,390 Waste production (Tn) 18,178 5,105 60,627 465,361 141 Public equipment (m2) 212,355 139,881 229,194 1,362,507 13,715 Street lamps (n) 2,323 1,821 1,850 8,166 249 Length of the water distribution network (m) 81,747 58,740 65,679 286,530 9,227

Surface area of cemeteries (m2) 6,293 4,105 7,770 57,012 760 Source: The authors, from data published by the Ministry of Finance and Public Administration.

The descriptive statistics for each year of the first-stage variables are shown in Annex 1.

Notably, the average total budget costs fell during the study period, as did the standard

deviation, i.e. the data had a lower level of dispersion. The total volume of waste decreased,

but all other output variables increased during the same period. In general, the standard

deviation of the data increased, except that for the total area of cemeteries.

The results for municipal efficiency are shown in Table 3.

Table 3. Cost efficiency of Canary Isles municipalities by population tranche (2008-

2012)

2008 2009 2010 2011 2012

Population (thousands) Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D. 50-20 0.9905 0.0403 0.9988 0.0050 1.0000 0.0000 0.9970 0.0136 0.9994 0.0270

20-10 1.0035 0.0068 0.9753 0.0779 0.9788 0.0732 1.0025 0.0150 0.9724 0.0658

10-5 1.0105 0.0663 1.0052 0.0495 0.9947 0.0695 1.0043 0.0438 0.9842 0.0702

<5 1.0306 0.1834 1.0445 0.0883 0.9948 0.0779 1.0302 0.0994 1.0322 0.0759

TOTAL 1.0097 0.1016 1.0072 0.0661 0.9930 0.0628 1.0084 0.0563 0.9997 0.0644

The above values show that, on average, these municipalities achieve a very high level of cost

efficiency. Those with fewer than 5,000 inhabitants obtain the highest values in this respect

during most of the period, while those with 10-20,000 inhabitants usually obtain the worst

Source: The authors, from data published by the Ministry of Finance and Public

Administration.Figure 1Municipal cost efficiency by population tranche (2008-2012)

21

results. Overall, average cost efficiency decreased during the study period, from 1.0097 to

0.9997. The pattern of cost efficiency varies according to the size of the population (see Fig.

1). Thus, in the larger municipalities, the cost efficiency increased slightly during the study

period, whilst it decreased in the others, with the exception of municipalities with fewer than

5,000 inhabitants, where the level of cost efficiency remained unchanged overall, although

with marked variations in the intermediate periods. In general, a worsening of the average

cost efficiency was perceived for all population tranches between 2008 and 2012.

Figure 1. Municipal cost efficiency by population tranche (2008-2012)

The descriptive statistics for the second-stage variables and the matrix of correlations are

presented in Annexes 2 and 3, respectively. There is no collinearity, and so the variance

inflation factor (VIF) was determined. According to the empirical rule proposed by Kutner et

al. (2004), problems of collinearity arise when VIF >10. In the present case, the average value

obtained was 3.24. None of the dependent variables obtained high values (the highest was

6.98), from which it can be concluded that there is no correlation between the explanatory

variables of the model. In addition, the standard deviations obtained show that, in general, the

data do not present a large degree of dispersion, except for per capita fiscal pressure and for

population density.

Table 4 shows the results of the truncated regression with panel data, performed to test the

study hypotheses, obtained using each of the above individual estimates as an independent

variable.

budget costs fell during the study period, as did the stand-ard deviation, i.e. the data had a lower level of dispersion.The total volume of waste decreased, but all other outputvariables increased during the same period. In general, thestandard deviation of the data increased, except that for thetotal area of cemeteries.

The results for municipal efficiency are shown in Table 3.The above values show that, on average, these municip-

alities achieve a very high level of cost efficiency. Thosewith fewer than 5,000 inhabitants obtain the highest valuesin this respect during most of the period, while those with10-20,000 inhabitants usually obtain the worst results. Over-all, average cost efficiency decreased during the study period,from 1.0097 to 0.9997. The pattern of cost efficiency variesaccording to the size of the population (see Fig. 1). Thus, inthe larger municipalities, the cost efficiency increased slightlyduring the study period, whilst it decreased in the others,with the exception of municipalities with fewer than 5,000inhabitants, where the level of cost efficiency remained un-changed overall, although with marked variations in the in-termediate periods. In general, a worsening of the aver-age cost efficiency was perceived for all population tranchesbetween 2008 and 2012.

The descriptive statistics for the second-stage variables andthe matrix of correlations are presented in Annexes 2 and 3,respectively. There is no collinearity, and so the variance in-flation factor (VIF) was determined. According to the empir-ical rule proposed by Kutner et al. (2004), problems of collin-earity arise when VIF >10. In the present case, the averagevalue obtained was 3.24. None of the dependent variablesobtained high values (the highest was 6.98), from which itcan be concluded that there is no correlation between the ex-planatory variables of the model. In addition, the standarddeviations obtained show that, in general, the data do notpresent a large degree of dispersion, except for per capitafiscal pressure and for population density.

Table 4 shows the results of the truncated regression withpanel data, performed to test the study hypotheses, obtainedusing each of the above individual estimates as an independ-ent variable.


Table 3Cost efficiency of Canary Isles municipalities by population tranche 2008-2012

20

4. Analysis of results: relationship between cost efficiency, financial condition and debt

This section presents the results obtained in the first and second stages, which are used to test

the research hypotheses.

Table 2 shows the descriptive statistics of the variables included in the first stage of the

analysis.

Table 2. Descriptive statistics: input and output variables (2008-2012).

Variable Mean Median Standard deviation Maximum Minimum

Total cost (euros) 13,388,847 8,643,118 12,451,851 77,448,391 1,279,124 No. of inhabitants (n) 12,865 8,605 11,085 44,007 772 Length of street network (m) 113,097 92,895 73,042 309,058 11,390 Waste production (Tn) 18,178 5,105 60,627 465,361 141 Public equipment (m2) 212,355 139,881 229,194 1,362,507 13,715 Street lamps (n) 2,323 1,821 1,850 8,166 249 Length of the water distribution network (m) 81,747 58,740 65,679 286,530 9,227

Surface area of cemeteries (m2) 6,293 4,105 7,770 57,012 760 Source: The authors, from data published by the Ministry of Finance and Public Administration.

The descriptive statistics for each year of the first-stage variables are shown in Annex 1.

Notably, the average total budget costs fell during the study period, as did the standard

deviation, i.e. the data had a lower level of dispersion. The total volume of waste decreased,

but all other output variables increased during the same period. In general, the standard

deviation of the data increased, except that for the total area of cemeteries.

The results for municipal efficiency are shown in Table 3.

Table 3. Cost efficiency of Canary Isles municipalities by population tranche (2008-

2012)

2008 2009 2010 2011 2012

Population (thousands) Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D. 50-20 0.9905 0.0403 0.9988 0.0050 1.0000 0.0000 0.9970 0.0136 0.9994 0.0270

20-10 1.0035 0.0068 0.9753 0.0779 0.9788 0.0732 1.0025 0.0150 0.9724 0.0658

10-5 1.0105 0.0663 1.0052 0.0495 0.9947 0.0695 1.0043 0.0438 0.9842 0.0702

<5 1.0306 0.1834 1.0445 0.0883 0.9948 0.0779 1.0302 0.0994 1.0322 0.0759

TOTAL 1.0097 0.1016 1.0072 0.0661 0.9930 0.0628 1.0084 0.0563 0.9997 0.0644

The above values show that, on average, these municipalities achieve a very high level of cost

efficiency. Those with fewer than 5,000 inhabitants obtain the highest values in this respect

during most of the period, while those with 10-20,000 inhabitants usually obtain the worst

Table 4Empirical results of the estimation of the determinants of municipal costefficiency

23

Variable Expected sign Value

Index of budget sustainability - -0.1135*** Level of government transfers received

+/- -0.0842

Net savings + 0.0007

Fiscal pressure per capita - -0.0001*

Short-term borrowing - -0.0232**

Long-term borrowing + -0.0013

Financial debt (short and long term)

- -0.0013

Commercial debt + 0.0076*

Compliance with borrowing limits

- -0.0718**

Index of average quality +/- 0.0906*

Political orientation of municipal government

+ -0.0450***

Type of municipal government +/- 0.0162

Population tranche +/- -0.0039

Population density +/- 0.0008*

Population aged under 15 years +/- -0.8607

Population aged over 65 years +/- 2.0917***

Index of unemployment +/- 0.5207

Index of economic activity +/- -0.0515*

Index of tourist activity - 0.0015

Dummy: 2009 -0.0331**

Dummy: 2010 -0.0474***

Dummy: 2011 -0.0511**

Dummy: 2012 -0.0930***

Constant 1.2797***

* p<0.05; ** p<0.01; *** p<0.001

Index of robustness: 73.05%

The first analysis performed of these results concerned the influence of the financial variables

on cost efficiency. The index of budget sustainability presents a negative sign and is

statistically significant (p<0.001). Consequently, Hypothesis 1 is verified, corroborating the

results obtained by Balaguer-Coll (2007), meaning that the higher the municipality’s costs in

relation to its income, the lower the cost efficiency. On the other hand, no significant evidence

was obtained for a relation between the variable reflecting vulnerability – the level of

transfers – and cost efficiency, unlike Geys (2006), De Borger and Kerstens (1996), Balaguer-

Coll (2002), Balaguer-Coll et al. (2007), Giménez and Prior (2007), Balaguer-Coll and Prior

(2009) and Bosch et al. (2012), who reported finding a significant relation between this

variable and cost efficiency. Accordingly, Hypothesis 2 cannot be accepted.

The first analysis performed of these results concerned theinfluence of the financial variables on cost efficiency. The in-dex of budget sustainability presents a negative sign and isstatistically significant (p<0.001). Consequently, Hypothesis1 is verified, corroborating the results obtained by Balaguer-Coll (2007), meaning that the higher the municipality’s costsin relation to its income, the lower the cost efficiency. Onthe other hand, no significant evidence was obtained for arelation between the variable reflecting vulnerability – thelevel of transfers – and cost efficiency, unlike Geys (2006), DeBorger and Kerstens (1996), Balaguer-Coll (2002), Balaguer-Coll et al. (2007), Giménez and Prior (2007), Balaguer-Colland Prior (2009) and Bosch et al. (2012), who reported find-ing a significant relation between this variable and cost effi-ciency. Accordingly, Hypothesis 2 cannot be accepted.

Of the variables measuring flexibility, those analysed wererelated to net savings, per capita fiscal pressure, debt andcompliance with the credit limit. In contrast to the findingsof Geys (2006) and Pérez-López et al. (2014), net savingsdid not present a significant relationship with cost efficiency,and so Hypothesis 3a cannot be accepted. On the other hand,fiscal pressure per capita presented a significant negative re-lationship (p<0.05) with cost efficiency, which supports Hy-pothesis 3b, and therefore we conclude that if the municipalgovernment lost the power to generate current income itwould be less motivated to manage its resources efficiently.These results coincide with those obtained by De Borger etal. (1994), Balaguer-Coll (2002), Giménez and Prior (2007),Balaguer-Coll et al. (2007), Balaguer-Coll and Prior (2009),Bosch et al. (2012) and Cuadrado-Ballesteros et al. (2013).

In relation to debt, in its various forms, which is themain object of this study, the variable short-term debtpresents a significant negative relationship with cost effi-ciency (p<0.01), meaning that higher financial costs, dueto higher levels of short-term debt, would reduce cost effi-ciency (Hypothesis 4a). These results coincide with thosereported by Worthington (2000) and Geys (2006). In con-trast, long-term debt does not present any significant rela-tionship with cost efficiency (Hypothesis 4b), which suggeststhat long-term debt does not affect short-term managementactivities, since most of the budget values used to calculatecost efficiency correspond to current expenditure (Mugera& Nyambane, 2015). This finding contrasts with Benitoet al. (2010), who observed a significant positive relation-ship between long-term debt and cost efficiency. The resultsfor commercial debt reveal a significant positive relationshipwith cost efficiency (p<0.05). Thus, for the first time thereis shown to be a clear difference between commercial andfinancial debt in relation to municipal cost efficiency (andtherefore Hypothesis 4c is accepted). This outcome could beinterpreted as meaning that financial debt affects cost effi-ciency due to the impact produced on the municipal budget,unlike commercial debt, which does not impose greater costson the municipality.

The indicator of compliance with the credit limit presentsa significant negative relationship with cost efficiency(p<0.01). Thus, the higher the credit limit, the fewer re-sources will be available to the municipality for public ser-vices and so the less cost efficient it will be (therefore, Hypo-thesis 5 is accepted).

In the second block of variables, we first considered the av-erage quality of public services and facilities, obtaining evid-ence of a significant positive relationship with municipal costefficiency (p<0.05), thus corroborating Zafra-Gómez andMuñiz (2010). In a previous study, Prior (2006) explainedthis relationship as the outcome of Total Quality Manage-ment, i.e. that investment in quality produces a correspond-ing increase in cost efficiency.

In line with De Borge et al. (1994), Benito et al. (2010),Geys et al. (2010) and Borge et al. (2008), we find that polit-


ical orientation is inversely associated with cost efficiency,and that the relation is statistically significant (p<0.001).However, no evidence was obtained that the type of mu-nicipal government affects municipal cost efficiency. Thesefindings coincide with Benito et al. (2010) and Pérez-Lópezet al. (2015), according to whom municipalities that are gov-erned by progressive political parties obtain higher levels ofcost efficiency.

In relation to the socioeconomic variables, our results con-firm, with different levels of significance, that greater popu-lation density is associated with higher levels of local gov-ernment cost efficiency. There is also significant evidencethat when the municipal population contains a higher pro-portion of older inhabitants, government cost efficiency isgreater, which corroborates the results obtained by Bosch etal. (2012).

Finally, the index of economic activity has a significant in-verse relationship with cost efficiency. However, no signific-ant relationships were found with respect to unemployment,the index of activity in the tourism sector, the size of the localpopulation aged under 15 years or the total population of themunicipality.

Conclusions and discussion

The recent economic crisis has heightened the need to im-prove the allocation of resources and to limit public sectorborrowing (in accordance with new regulations in this re-spect). This situation has spurred research interest in eval-uating how well municipalities manage their affairs and indetermining which factors have most influence on cost effi-ciency in this sector, especially those related to sources offinance and types of municipal debt.

The theoretical arguments considered in previous researchindicate that, in general, public sector borrowing is inverselyassociated with cost efficiency, due (at least in part) to thehigher spending and lower cost efficiency resulting fromincreased financial costs (Worthington, 2000; Geys, 2006;Benito el al., 2010; Da Cruz & Marques, 2014; Cordero etal., 2017). In this context, we apply an order-m methodo-logy (thus overcoming the limitations of DEA), followed bybootstrap truncated regression with panel data (as proposedby Simar and Wilson, 2007). The novel aspect of the presentstudy is its emphasis on the need to distinguish between com-mercial and financial borrowing, having detected that the re-lationship between municipal debt and cost efficiency variesaccording to the type of debt incurred. We find that an in-crease in commercial debt is associated with greater cost ef-ficiency, possibly derived from the absence of financial costsin this type of borrowing. Financial debt, on the other hand,impairs municipal cost efficiency. These results highlight theneed for further study of the relationship between commer-cial debt and cost efficiency, taking into account that thisrelationship has been detected in municipalities of the Ca-nary Isles (Spain) during the period of financial crisis, andshould be corroborated by reference to other socioeconomiccontexts.

Confirming previous research, we also detected a relation-ship between cost efficiency and other components of finan-cial condition, namely budget sustainability, fiscal pressureand borrowing limits. But, in addition, we show that thisrelationship exists in periods of transboundary crisis.

The results obtained show that, in times of crisis, progress-ive governments obtain better levels of cost efficiency thanconservative ones, that greater service quality improves cost

efficiency and that certain socioeconomic variables impact onthe cost efficiency achieved.

The implications drawn from the present study are clearand important from the standpoint of local government man-agement and financing. In a context of reduced spending,characterised by the presence of a transboundary crisis, an in-crease in commercial debt can bring about efficiency improve-ments in municipalities where the tourism industry plays animportant role, while resorting to short-term financial debtworsens cost efficiency. Municipal managers in such areasshould take these circumstances into account when consider-ing their financing options during periods of financial reces-sion.

A limitation of the present study, but also an indication ofan area of interest for future research, is the need to expandthe study to address other regions, in order to examine ourhypotheses of cost efficiency vs. commercial and financialdebt in other municipal contexts. Furthermore, other meth-ods of analysis should be considered, such as the DEA-datapanel approach used by Surroca et al. (2016), which takesinto account the unobservable heterogeneity in the units eval-uated, or the DEA-data panel models developed by Pérez-López et al. (2018) for determining scale efficiency.

Acknowledgements

We are grateful for the financial support received from Min-isterio de Educación y Ciencia (Spain; ECO2013-48413-R,ECO2016-76578-R and ECO2017-88241-R).

Conflict of interests

The authors declare no conflict of interests.

References

Ansell, C., Boin, A., & Keller, A. (2010). Managing trans-boundary crises: identifying the building blocks of an effect-ive response system. Journal of Contingencies and CrisisManagement, 18(4), 195–207.

Andrews, R., & Entwistle, T. (2013). Four faces of publicservice efficiency: What, how, when and for whom to pro-duce. Public Management Review, 15(2), 246-264.

Atan, R., Raman, S. A., Sawiran, M. S., Mohamed, N., &Mail, R. (2010). Financial Performance of Malaysian LocalAuthorities: A Trend Analysis. In International Conferenceon Science and Social Research, 271–276. Kuala Lumpur,December 5–8. CSSR.

Balaguer-Coll, M. T. (2002). Análisis de la situación finan-ciera y la eficiencia en las administraciones locales. Institutode Contabilidad y de Auditoría de Cuentas.

Balaguer-Coll, M. T. (2004). La eficiencia en las adminis-traciones locales ante diferentes especificaciones del output.Hacienda Pública Española, 170, 37-58.

Balaguer-Coll, M. T., & Prior, D. (2009). Short-and long-term evaluation of efficiency and quality. An application toSpanish municipalities. Applied Economics, 41(23), 2991-3002.

Balaguer-Coll, M. T., Prior, D., & Tortosa-Ausina, E. (2007).On the determinants of local government performance: Atwo-stage nonparametric approach. European Economic Re-view, 51(2), 425-451.

Balaguer-Coll, M. T., Prior, D., & Tortosa-Ausina, E. (2013).Output complexity, environmental conditions, and the effi-


ciency of municipalities. Journal of Productivity Analysis,39(3), 303-324.

Bastida, F. J., Benito, B., & Guillamón, M. D. (2009). Anempirical assessment of the municipal financial situation inSpain. International Public Management Journal, 12(4),484-499.

Bastida, F., Beyaert, A., & Benito, B. (2013). Electoralcycles and local government debt management. Local Gov-ernment Studies, 39(1), 107-132.

Benito, B., & Bastida, F. J. (2004). The determinants of themunicipal debt policy in Spain. Journal of Public Budgeting,Accounting and Financial Management, 16(4), 525-558.

Benito, B., Guillamón, M.D., Ríos, A.M., & Bastida, F.(2018). Can salaries and re-election prevent political cor-ruption? An empirical evidence, Revista de Contabilidad-Spanish Accounting Review, 21 (1), 19-27.

Benito, B., Bastida, F. J., & García, J. A. (2010). Explainingdifferences in efficiency: An application to Spanish municip-alities. Applied Economics, 42(4), 515-528.

Bisogno, M., & Cuadrado-Ballesteros, B. (2018). Explain-ing the relation between efficiency and financial condition:empirical analysis on local governments. In Financial Sus-tainability and Intergenerational Equity in Local Govern-ments (pp. 149-178). IGI Global.

Borge, L., Falch, T., & Tovmo, P. (2008). Public sector effi-ciency: The roles of political and budgetary institutions, fiscalcapacity, and democratic participation. Public Choice, 136(3-4), 475-495.

Borowiecki, K. J. (2013). Agglomeration economies inclassical music. Papers in Regional Science, in press. https://doi.org/10.1111/pirs.12078.

Bosch, N., Espasa, M., & Mora, T. (2012). Citizen controland the efficiency of local public services. Environment andPlanning-Part C, 30(2), 248.

Bruns, C., & Himmler, O. (2011). Newspaper circulationand local government efficiency. The Scandinavian Journalof Economics, 113(2), 470-492.

Cabaleiro, R., Buch, E., & Vaamonde, A. (2013). Develop-ing a method to assessing the municipal financial health. TheAmerican Review of Public Administration, 43(6), 729-751.

Carmeli, A. (2002). A conceptual and practical frameworkof measuring performance of local authorities in financialterms: analysing the case of Israel. Local Government Stud-ies 28 (1): 21–36.

Cazals, C., Florens, J., & Simar, L. (2002). Nonparametricfrontier estimation: A robust approach. Journal of Econo-metrics, 106(1), 1-25.

Canadian Institute of Chartered Accountants. (1997). In-dicators of government financial condition. Toronto, Ontario,Canada: CICA.

Canadian Institute of Chartered Accountants. (2009).Public sector statements of recommended practice (SORP) 4.Indicators of financial condition. Toronto, Ontario, Canada:CICA.

Cabasés, F., P. Pascual y J. Valles (2007), The Effectivenessof Institutional Borrowing Restrictions: Empirical Evidencefrom Spanish Municipalities, Public Choice, 131 (3), pp. 293-313.

Congleton, R. (2002). The Median Voter Model. In CKRowley (Ed.); F. Schneider (Ed.) (2003). The Encyclopediaof Public Choice.

Cordero, J.M., Pedraja-Chaparro, F., Pisaflores, E. C.,& Polo, C. (2017). Efficiency assessment of Portuguesemunicipalities using a conditional nonparametric approach.Journal of Productivity Analysis, 48(1), 1-24.

Cordero, J. M., Prior, D., & Simancas, R. (2013). A com-

parison of public and private schools in Spain using robustnonparametric frontier methods. Central European Journalof Operations Research, 1-22.

Cuadrado-Ballesteros, B., & Bisogno, M. (2018). Efficiencyas a determinant of financial condition: an assessment ofItalian and Spanish local governments. International PublicManagement Journal, (just-accepted), 1-58.

Cuadrado-Ballesteros, B., García-Sánchez, I. M., & Prado-Lorenzo, J. M. l. (2013). Effect of modes of public servicesdelivery on the efficiency of local governments: A two-stageapproach. Utilities Policy, 26, 23-35.

Da Cruz, N. F., & Marques, R. C. (2014). Revisiting thedeterminants of local government performance. Omega, 44,91-103.

Daraio, C., & Simar, L. (2007). Advanced robust and non-parametric methods in efficiency analysis: Methodology andapplications. Springer Science & Business Media.

De Borger, B., & Kerstens, K. (1996). Cost efficiency of Bel-gian local governments: A comparative analysis of FDH, DEA,and econometric approaches. Regional Science and UrbanEconomics, 26(2), 145-170.

De Borger, B., Kerstens, K., Moesen, W., & Vanneste, J.(1994). Explaining differences in productive efficiency: Anapplication to Belgian municipalities. Public Choice, 80(3-4),339-358.

De Sousa, Maria Da Conceição Sampaio, & Stoi, B. (2005).Technical efficiency of the brazilian municipalities: Cor-recting nonparametric frontier measurements for outliers.Journal of Productivity Analysis, 24(2), 157-181.

De Witte, K., & Geys, B. (2013). Citizen coproduction andefficient public good provision: Theory and evidence fromlocal public libraries. European Journal of Operational Re-search, 224, 592-602.

De Witte, K., & Marques, R. C. (2010). Influential obser-vations in frontier models, a robust non-oriented approachto the water sector. Annals of Operations Research, 181(1),377-392.

Diez-Ticio, A., & Mancebón, M. (2002). The efficiency ofthe Spanish police service: An application of the multiactivityDEA model. Applied Economics, 34(3), 351-362.

Dijkgraaf, E., Gradus, R.H.J.M., & Melenberg, B. (2003).Contracting out refuse collection. Empirical Economics, 28(3), 553–570.

Dubin, J. A., & Navarro, P. (1988). How markets for im-pure public goods organize: The case of household refusecollection. Journal of Law, Economics, & Organization, , 217-241.

Fernández Santos, Y., Martínez Campillo, A., & FernándezFernández, J. M. (2015). Determinantes de la eficiencia enlas universidades públicas españolas: Un análisis DEA en dosetapas. XIV Jornadas De Economía De La Educación. Madrid.Asociación De Economía De La Educación,

Fresneda, S., & Hernández, J. (2018). Institutional capa-city in the accounting reform process in Spanish local govern-ments. Revista de Contabilidad – Spanish Accounting Review,21 (2), 188–195, https://doi.org/10.1016/j.rcsar.2018.05.002

Geys, B. (2006). Looking across borders: A test of spatialpolicy interdependence using local government efficiency rat-ings. Journal of Urban Economics, 60(3), 443-462.

Geys, B., Heinemann, F., & Kalb, A. (2010). Voter involve-ment, fiscal autonomy and public sector efficiency: Evidencefrom German municipalities. European Journal of PoliticalEconomy, 26(2), 265-278.

Giménez, V. M., & Prior J., D. (2007). Longand shorttermcost efficiency frontier evaluation: Evidence from Spanish

https://doi.org/10.1111/pirs.12078

https://doi.org/10.1111/pirs.12078

https://doi.org/10.1016/j.rcsar.2018.05.002

https://doi.org/10.1016/j.rcsar.2018.05.002


local governments. Fiscal Studies, 28(1), 121-139.Government of Western Australia: Department of Local

Government. 2013. Local Government OperationalGuidelines: Financial Ratios, 18. Perth: Department of LocalGovernment.

Greenberg, J., & Hillier, D. (1995). Indicators of finan-cial condition for governments. 5th Conference of Comparat-ive International Governmental Accounting Research. Paris-Amy.

Groves, M., Godsey, W., & Shulman, M. (2003). Evaluat-ing Financial Condition: A Handbook of Local Government(3rd ed.). Washington, DC: International City ManagementAssociation.

Hammer, R. B., & Green, G. P. (1996). Local growth promo-tion: Policy adoption versus effort. Economic DevelopmentQuarterly, 10(4), 331-341.

Hendrick, R. M. (2011). Managing the fiscal metropolis:The financial policies, practices, and health of suburban mu-nicipalities Georgetown University Press.

Ibarloza Arrizabalaga, A., Erkizia Olaizola, A., MallesFernández, E. (2017). Implantación de una metodología co-operativa de cálculo de costes de los servicios públicos locales.Revista de Contabilidad-Spanish Accounting Review, 20 (2),210-224.

Kalb, A., Geys, B., & Heinemann, F. (2012). Value formoney? German local government efficiency in a compar-ative perspective. Applied Economics, 44(2), 201-218.

Kneip, A., Park, B. U., & Simar, L. (1998). A note on theconvergence of nonparametric DEA estimators for produc-tion efficiency scores. Econometric Theory, 14(06), 783-793.

Kutner, M. H., Nachtsheim, C., & Neter, J. (2004). Appliedlinear regression models. McGraw-Hill/Irwin.

López-Hernández, A., Prior J., D., Zafra-Gómez, J. L.,&Plata Díaz, A. M. (2012). Determinando el impacto delentorno socioeconómico y de las formas de gestión sobre laeficiencia en la prestación de servicios públicos locales de An-dalucía. Fundación Pública Andaluza Centro de Estudios An-daluces. Junta de Andalucía.

López-Hernández, A.M., Zafra-Gómez, J. L., Plata Díaz, A.M., & de la Higuera-Molina, E. J. (2017). Modeling fiscalstress and contracting out in local government: The influenceof time, financial condition, and the great recession. TheAmerican Review of Public Administration, 48 (6), 565-583.https://doi.org/10.1177/0275074017699276.

Mugera, A. W., & Nyambane, G. G. (2015). Impact of debtstructure on production efficiency and financial performanceof broadacre farms in Western Australia. Australian Journalof Agricultural and Resource Economics, 59(2), 208-224.

NarbónPerpiñá, I., & De Witte, K. (2018a). Local govern-ments’ efficiency: a systematic literature review—part I. In-ternational Transactions in Operational Research, 25(2), 431-468.

NarbónPerpiñá, I., & De Witte, K. (2018b). Local gov-ernments’ efficiency: a systematic literature review—part II.International Transactions in Operational Research, 25(2),1107-1136.

Pascual Arzoz, P., Cabasés Hita, F., & Ezcurra, R. (2008).Financiación, restricciones institucionales y endeudamiento:Un análisis con microdatos de los municipios de Navarra. XVEncuentro de Economía Pública.

Pérez-López, G., Plata-Díaz, A. M., Zafra-Gómez, J. L., &López-Hernández, A. M. (2014). Operaciones fuera de pre-supuesto (off budget), factores políticos y deuda municipal:Un estudio empírico aplicando una metodología de datos depanel. Gestión y Política Pública, 23(1), 185-218.

Pérez-López, G., Prior, D., & Zafra-Gómez, J. L. (2015).

Rethinking new public management delivery forms and ef-ficiency: long-term effects in Spanish local government.Journal of Public Administration Research and Theory, 25(4),1157-1183.

Pérez-López, G., Prior, D., Zafra-Gómez, J.L., & Plata-Díaz,A.M. (2016). Cost efficiency in municipal solid waste servicedelivery. Alternative management forms in relation to localpopulation size. European Journal of Operational Research,255, 583-592. doi:10.1016/j.ejor.2016.05.034.

Pérez-López, G., Prior, D., & Zafra-Gómez, J. L. (2018).Temporal scale efficiency in DEA panel data estimations.An application to the solid waste disposal service in Spain.Omega,76, 18-27.

Plata-Díaz, A. M., de la Higuera-Molina, E. J., Garrido-Rodríguez, J. C., & Zafra-Gómez, J. L. (2017). Con-tracting out and social services: responses to the auster-ity machine, financial condition and political parties. Ad-ministration & Society, in press, https://doi.org/10.1177/0095399717713797.

PricewaterhouseCoopers. 2006. National Financial Sus-tainability Study of Local Government: Commissioned bythe Australian Local Government Association. Sydney, NSW:LGA.

Prieto, A. M., & Zofio, J. L. (2001). Evaluating effective-ness in public provision of infrastructure and equipment: thecase of Spanish municipalities. Journal of productivity Ana-lysis, 15(1), 41-58.

Prior, D. (2006). Efficiency and total quality managementin health care organizations: A dynamic frontier approach.Annals of Operations Research, 145(1), 281-299.

Prior, D., Surroca, J., & Tribó, J. A. (2008). Are sociallyresponsible managers really ethical? Exploring the relation-ship between earnings management and corporate social re-sponsibility. Corporate Governance: An International Review,16(3), 160-177.

Simar, L. (2003). Detecting outliers in frontier models:A simple approach. Journal of Productivity Analysis, 20(3),391-424.

Simar, L., & Wilson, P. W. (2007). Estimation and infer-ence in two-stage, semi-parametric models of production pro-cesses. Journal of Econometrics, 136(1), 31-64.

Sung, N. (2007). Information technology, efficiency andproductivity: Evidence from korean local governments. Ap-plied Economics, 39(13), 1691-1703.

Tiebout, C. M. (1956). A pure theory of local expenditures.Journal of Political Economy, 64(5), 416-424.

Turley, G., Robbins, G., & McNena, S. (2015). A frameworkto measure the financial performance of local governments.Local Government Studies, 41(3), 401-420.

Vanden Eeckaut, P., Tulkens, H., & Jamar, M. (1993). Costefficiency in Belgian municipalities. The Measurement of Pro-ductive Efficiency–Techniques and Applications, 300-334.

Wang, X., Dennis, L., & Sen,Y. (2007). Measuring finan-cial condition: A study of U.S. states. Public Budgeting andFinance, 27(2), 1-21.

Worthington, A. C. (2000). Cost efficiency in Australianlocal government: A comparative analysis of mathematicalprogramming and econometrical approaches. Financial Ac-countability & Management, 16(3), 201-223.

Zafra-Gómez, J. L., Hernández Bastida, A., & López-Hernández, A. (2006). Evaluación de la condición financi-era en las administraciones locales a través del análisis decomponentes principales. Presupuesto y Gasto Público, 43,113-126.

Zafra-Gómez, J. L., & López-Hernández, A. (2006). Lavaloración de la condición financiera en los ayuntamien-

https://doi.org/10.1177/0275074017699276

doi:10.1016/j.ejor.2016.05.034

https://doi.org/10.1177/0095399717713797

https://doi.org/10.1177/0095399717713797


tos andaluces. Auditoría Pública: Revista de los ÓrganosAutónomos de Control Externo, (39), 41-52.

Zafra-Gómez, J. L., López-Hernández, A., & Hernández-Bastida, A. (2009a). Developing a model to measure finan-cial condition in local government evaluating service qual-ity and minimizing the effects of the socioeconomic environ-ment: An application to Spanish municipalities. The Amer-ican Review of Public Administration, 39(4), 425-449.

Zafra-Gómez, J. L., López-Hernández, A. M., & Hernández-Bastida, A. (2009b). Evaluating financial performance inlocal government: maximizing the benchmarking value. In-ternational Review of Administrative Sciences, 75(1), 151-167.

Zafra-Gómez, J. L., & Muñiz Pérez, A. (2010). Overcom-ing cost-inefficiencies within small municipalities: Improvefinancial condition or reduce the quality of public services?Environment and Planning.C, Government & Policy, 28(4),609-629.


Table Annex 1Descriptive statistics of input and output variables, by year

33

Annex 1. Descriptive statistics of input and output variables, by year

2008 2009 2010 2011 2012 Mean S.D. Mean S.D. Mean S.D. Mean S.D. Mean S.D. Total cost (thousands of euros) 13,730 11,762 14,640 12,970 13,569 12,710 13,558 13,191 11,448 11,626 No. of inhabitants (n) 12,559 10,704 12,787 10,959 12,894 11,100 13,015 11,245 13,069 11,416 Length of street network (m) 99,367 62,930 115,435 75,246 118,288 76,857 115,935 75,020 116,459 75,160 Waste production (Tn) 12,032 15,099 8,948 10,922 8,884 10,647 10,486 18,161 10,541 18,305 Public equipment (m2) 201,642 207,599 211,201 241,476 215,825 233,415 213,476 230,256 219,632 233,222 Street lamps (n) 2,209 1,730 2,329 1,907 2,356 1,867 2,401 1,902 2,318 1,843 Length of the water distribution network (m)

77,732 60,586 78,497 61,278 84,947 69,281 84,164 68,013 83,396 69,234

Surface area of cemeteries (m2) 6,191 7,887 6,244 7,857 6,313 7,702 6,316 7,702 6,399 7,702

Table Annex 2Descriptive statistics of the second-stage variables and variance inflation factor (VIF)

34

Annex 2. Descriptive statistics of the second-stage variables and variance inflation factor

(VIF)

VARIABLE MEAN MEDIAN STANDARD DEVIATION MINIMUM MAXIMUM VIF

MUNICIPAL EFFICIENCY 1.003599 1 0.071952 0.537322 1.557485 INDEX OF BUDGET SUSTAINABILITY 0.969930 0.958446 0.139215 0.523109 1.619950 1.4 LEVEL OF GOVERNMENT TRANSFERS RECEIVED

0.588158 0.600236 0.174148 0.160724 0.959501 5.35

SHORT-TERM BORROWING 14.880690 14.908130 1.186554 11.479450 17.366010 3.72 LONG-TERM BORROWING 14.563010 14.717250 1.653298 7.453046 17.515700 3.16 FINANCIAL DEBT 14.745870 15.004000 1.692731 7.071955 17.639760 2.24 COMMERCIAL DEBT 14.630770 14.656460 1.226376 6.057557 17.060300 2.36 COMPLIANCE WITH BORROWING LIMITS 0.385593 0.336378 0.304140 0 1.798995 1.89 NET SAVINGS 13.699520 13.696310 1.388470 5.895036 16.641880 2.45 FISCAL PRESSURE PER CAPITA 431.932200 346.530000 246.656500 102.07000 1491.210000 6.98 INDEX OF SERVICE QUALITY 2.435642 2.448859 0.160990 1.781336 2.759871 1.27 POITICAL ORIENTATION 0.163636 0 0.370427 0 1 1.22 TYPE OF MUNICIPAL GOVERNMENT 0.420779 0 0.494327 0 1 1.36 POPULATION TRANCHE 0.989610 1 0.710704 0 2 4.56 INDEX OF UNEMPLOYMENT 0.107202 0.107918 0.026992 0.043742 0.174494 1.84 INDEX OF ECONOMIC ACTIVITY 0.983667 0.671893 1.101559 0 10.057220 1.47 INDEX OF TOURISM ACTIVITY 7.175759 0.785855 14.375890 0 66.829270 4.39 POPULATION AGED UNDER 15 YEARS 0.135765 0.139684 0.026380 0.060270 0.188756 7.02 POPULATION AGED OVER 65 YEARS 0.165357 0.161954 0.058262 0.056045 0.315292 7.3 POPULATION DENSITY 291.988100 179.199500 454.809300 6.561173 3759.106000 1.49 MEAN VIF 3.24


Table Annex 3Correlation matrix of study variables

1

N

et s

avin

gs

Fina

ncia

l de

bt

Com

mer

cial

de

bt

Long

-term

bo

rrow

ing

Shor

t-ter

m

borr

owin

g

Com

plia

nce

with

bo

rrow

ing

limits

Leve

l of

gove

rnm

ent

tran

sfer

s re

ceiv

ed

Fisc

al

pres

sure

per

ca

pita

Inde

x of

bu

dget

su

stai

nabi

lity

Inde

x of

se

rvic

e qu

ality

Polit

ical

or

ient

atio

n

Type

of

mun

icip

al

gove

rnm

ent Po

pula

tion

tran

che

Inde

x of

un

empl

oym

ent

Inde

x of

ec

onom

ic

activ

ity

Inde

x of

to

uris

m

activ

ity

Popu

latio

n ag

ed u

nder

15

yea

rs

Popu

latio

n ag

ed o

ver

65

year

s

Popu

latio

n de

nsity

Net

sav

ings

1.0

000

Fina

ncia

l deb

t 0.0

457

1.000

0

Com

mer

cial

de

bt

0.106

4*

0.688

3***

1.0

000

Long

-term

bo

rrow

ing

0.449

3***

-0

.0141

0.0

366

1.000

0

Shor

t-ter

m

borr

owin

g 0.5

179*

**

-0.03

41

0.054

2 0.6

983*

**

1.000

0

Co

mpl

ianc

e w

ith b

orro

win

g lim

its

0.035

2 -0

.0105

0.0

567

0.521

8***

0.3

995*

**

1.000

0

Leve

l of

gove

rnm

ent

tran

sfer

s re

ceiv

ed

-0.59

56**

* 0.0

234

-0.01

65

-0.61

13**

* -0

.5615

***

-0.26

35**

* 1.0

000

Fisc

al p

ress

ure

per

capi

ta

0.555

8***

0.0

057

0.045

6 0.4

374*

**

0.425

7***

0.1

500*

**

-0.79

08**

* 1.0

000

In

dex

of

budg

et

sust

aina

bilit

y -0

.3370

***

-0.07

37

-0.02

67

-0.09

21*

0.020

4 0.0

276

0.119

6**

-0.26

70**

* 1.0

000

Inde

x of

se

rvic

e qu

ality

0.1

540*

**

-0.04

03

-0.03

94

0.179

4***

0.2

049*

**

0.020

7 -0

.1080

**

0.064

9 0.0

106

1.000

0

Polit

ical

or

ient

atio

n -0

.0405

-0

.0144

-0

.0586

0.0

139

0.007

7 -0

.1007

**

-0.03

65

-0.00

46

-0.00

68

0.166

2***

1.0

000

Type

of

mun

icip

al

gove

rnm

ent

0.126

6**

-0.14

95**

* -0

.0758

0.2

526*

**

0.275

0***

0.2

514*

**

-0.26

14**

* 0.1

577*

**

-0.02

96

0.010

8 0.0

639

1.000

0

Popu

latio

n tr

anch

e 0.5

816*

**

-0.04

25

-0.01

32

0.629

8***

0.7

267*

**

0.119

8**

-0.58

11**

* 0.3

672*

**

-0.02

49

0.333

0***

0.0

460

0.145

9***

1.0

000

Inde

x of

un

empl

oym

ent

0.014

1 -0

.1319

**

-0.20

23**

* -0

.0159

0.0

941*

-0

.1103

**

-0.01

94

-0.25

61**

* 0.0

979*

0.0

983*

0.0

502

0.072

3 0.3

031*

**

1.000

0

Inde

x of

ec

onom

ic

activ

ity

0.222

2***

0.0

082

0.000

7 0.1

741*

**

0.188

9***

-0

.0349

-0

.2657

***

0.310

8***

0.0

248

0.211

6***

0.2

437*

**

0.018

8 0.2

657*

**

0.029

6 1.0

000

Inde

x of

to

uris

m a

ctiv

ity

0.380

4***

0.0

032

0.053

1 0.3

466*

**

0.403

8***

0.1

79**

* -0

.5450

***

0.782

2***

-0

.1554

***

0.116

1**

-0.04

91

0.234

4***

0.2

542*

**

-0.35

60**

* 0.3

536*

**

1.000

0

Popu

latio

n ag

ed u

nder

15

year

s 0.4

542*

**

0.015

7 0.0

007

0.445

5***

0.5

613*

**

0.147

2***

-0

.5107

***

0.268

2***

0.0

120

0.148

6***

-0

.0491

0.1

612*

**

0.641

6***

0.2

558*

**

0.157

***

0.093

7*

1.000

0

Po

pula

tion

aged

ove

r 65

ye

ars

-0.54

83**

* -0

.0070

-0

.0133

-0

.4515

***

-0.59

75**

* -0

.1336

***

0.524

0***

-0

.3944

***

0.066

1 -0

.2574

***

0.051

9 -0

.1931

***

-0.65

21**

* -0

.1239

**

-0.21

28**

* -0

.2971

***

-0.86

79**

* 1.0

000

Popu

latio

n de

nsity

0.1

664*

**

-0.11

01**

0.0

136

0.217

4***

0.3

332*

**

0.014

6 -0

.2095

***

0.049

1 0.1

062*

* 0.2

498*

**

0.032

0 0.1

461*

**

0.388

0***

0.2

112*

**

0.141

9***

0.1

837*

**

0.081

1 -0

.0698

1.0

000

cost efﬁciency and ﬁnancial situation of local governments

Documents